[porsager]

There is quite a big difference, but I'll highlight some of the main points here. Note I'm the author of Postgres.js so I'm obviously biased.

Slonik is a wrapper around another node driver (pg), so it's performance is the same as that, where Postgres.js is significantly faster (2-5x)[1].

Postgres.js is also a zero dependency module, whereas Slonik has quite the dependency graph meaning - compare https://npmgraph.js.org/?q=slonik with https://npmgraph.js.org/?q=postgres. That makes it more difficult to audit the code and preventing chain attacks etc.

Slonik also doesn't have the same lean straight forward developer experience as Postgres.js - again I'm biased saying that ;)

Postgres.js also does things that will make your queries perform better out of the box by implicitly creating prepared statements and using pipelining.

I suppose you could build Slonik on top of Postgres.js instead of pg as well, but that would probably only make sense if migrating to Postgres.js.

[1] http://github.com/porsager/postgres-benchmarks#results

[sdoering]

> Postgres.js is also a zero dependency module, whereas Slonik has quite the dependency graph meaning - compare https://npmgraph.js.org/?q=slonik with https://npmgraph.js.org/?q=postgres.

This one just made my day. Thanks. I remember trying to build a tool at work with as little as possible dependencies (in python) and how satisfying it was to see quite a few dependencies just being wrappers replaces with 5 lines of my own code that i could easily audit and ensure no supply chain attack was possible for that functionality.